Question

Could you clarify some ideas behind Python classes and class instances?

Consider this:

class A():
    name = 'A'

a = A()

a.name = 'B' # point 1 (instance of class A is used here)

print a.name
print A.name

prints:

B
A

if instead in point 1 I use class name, output is different:

A.name = 'B' # point 1 (updated, class A itself is used here)

prints:

B
B

Even if classes in Python were some kind of prototype for class instances, I’d expect already created instances to remain intact, i.e. output like this:

A
B

Can you explain what is actually going on?

Answer 1

First of all, the right way in Python to create fields of an instance (rather than class fields) is using the __init__ method. I trust that you know that already.

Python does not limit you in assigning values to non-declared fields of an object. For example, consider the following code:

class Empty: pass
e = Empty()
e.f = 5
print e.f # shows 5

So what’s going in your code is:

1. You create the class A with a static field name assigned with A.
2. You create an instance of A, a.
3. You create a new field for the object a (but not for other instances of A) and assign B to it
4. You print the value of a.name, which is unique to the object a.
5. You print the value of the static field A.name, which belongs to the class